Working Paper Series Congressional Budget Office Washington, D.C. Inflation, Inflation Expectations, and the Phillips Curve Yiqun Gloria Chen Congressional Budget Office [email protected] Working Paper 2019-07 August 2019 To enhance the transparency of the work of the Congressional Budget Office and to encourage external review of that work, CBO’s working paper series includes papers that provide technical descriptions of official CBO analyses as well as papers that represent independent research by CBO analysts. Papers in this series are available at http://go.usa.gov/ULE. This paper has not been subject to CBO’s regular review and editing process. The views expressed here should not be interpreted as CBO’s. The author thanks the following CBO staff for their comments and suggestions: Robert Arnold, Wendy Edelberg, Kim Kowalewski (formerly of CBO), and Jeffrey Werling. The author also thanks Alan Detmeister of UBS, Jan Hatzius of Goldman Sachs, David Lebow of the Board of Governors of the Federal Reserve System, and James Stock of Harvard University for their useful feedback. www.cbo.gov/publication/55501 Abstract This paper studies the current state of inflation dynamics through the lens of the Phillips curve and assesses the degree of anchoring of inflation expectations. I first estimate a Phillips curve model with both past inflation and a constant anchor as explanatory variables over the 1999– 2018 period for a variety of measures of consumer prices. My results show that the Phillips curve has shifted away from an accelerationist form toward a level form, but that shift is incomplete, particularly for core inflation. I then turn to survey measures of professional forecasters’ and consumers’ inflation expectations and assess the degree to which those expectations are anchored. My analysis shows that although professional forecasters’ expectations have been well anchored, consumers’ expectations have not. Further analysis using multiple empirical measures of inflation expectations suggests that in the context of the Phillips curve, consumers’ expectations have generally outperformed professional forecasters’ expectations in terms of explaining and forecasting the dynamics of inflation over the past two decades. In addition to analyzing the form of inflation expectations in the Phillips curve model, this paper examines the slope of the Phillips curve, or the sensitivity of inflation to cyclical fluctuations in economic conditions. I follow Stock and Watson (2018) and estimate the Phillips curve for various components of aggregate inflation. The results show that whereas price changes for many service categories (particularly shelter) and food remain largely procyclical, price changes for most goods have been either noncyclical or countercyclical over the past two decades. That finding helps explain the flatness of the Phillips curve on the aggregate level as well as the variation in cyclical sensitivity among the different aggregate inflation measures. Keywords: inflation, Phillips curve, business cycle, inflation expectations, anchoring JEL Classification: E17, E31, E37 Contents 1. Introduction ................................................................................................................................. 1 2. Ten Measures of Inflation ........................................................................................................... 8 Overall Inflation ..................................................................................................................... 8 Core Inflation ......................................................................................................................... 9 3. Reestimating the Phillips Curve................................................................................................ 14 Benchmark Specifications ................................................................................................... 14 Estimation Results ............................................................................................................... 16 Cyclical Sensitivity: A Component-Level Analysis ............................................................ 21 4. Inflation Expectations ............................................................................................................... 23 Are Inflation Expectations Anchored?................................................................................. 24 Whose Inflation Expectations Matter?................................................................................. 34 5. Conclusion ................................................................................................................................ 36 Appendix A. Stability Checks....................................................................................................... 38 Appendix B. Inflation Expectations of Financial Market Participants ......................................... 40 Appendix C. The Phillips Curve With Survey Measures of Inflation Expectations .................... 41 References ..................................................................................................................................... 52 1. Introduction From the “great inflation” of the 1970s to the “missing disinflation” of 2009 to 2011, the dynamics of inflation in the United States has changed profoundly over the past six decades (see Figure 1). To a large extent, that evolution is both a cause and a consequence of the changing nature of inflation expectations over the same period (see Figure 2). This paper examines the current state of inflation through the lens of the reduced-form Phillips curve—a regression model that links the dynamics of inflation to inflation expectations and the state of the real economy— and assesses the current properties of inflation expectations. The degree of stability in inflation depends critically on the properties of inflation expectations, particularly, whether they are anchored or not. Because unanchored expectations tend to correlate heavily with past inflation (a phenomenon known as backward-looking or adaptive inflation expectations), they can serve as an “accelerator” for the effects of excess or shortfall in demand in the labor and product markets and for the effects of transitory shocks from the supply side of the economy—such as a sudden rise in the price of oil or a sudden change in the exchange value of the U.S. dollar—by forming a feedback loop between inflation expectations and actual inflation. As a result, inflation tends to rise or fall persistently when expectations are unanchored and the economy is in disequilibrium. By contrast, perfectly anchored inflation expectations do not respond to transitory shocks (that is, they are “shock anchored”) and can be tied to a particular level specified by a central bank that sets a target for inflation (a process known as level anchoring; see Ball and Mazumder, 2011). As a result, inflation tends to stay relatively stable when expectations are anchored. Inflation dynamics in the United States from the late 1960s through the mid-1990s are considered to be generally consistent with the characteristics of an “accelerationist” regime. Trend inflation, as measured by the five-year average rate of annual growth in the personal consumption price (PCE) price index, rose persistently from less than 2 percent in mid-1960s to over 8 percent in the late 1970s and early 1980s before falling gradually back to around 2 percent by the mid-1990s. Over the same period, empirical measures of inflation expectations also rose and fell persistently, generally tracking actual inflation but with a considerable lag. To model the inflation dynamics during that period, economists have used an “accelerationist” Phillips curve model in which inflation expectations are modeled by distributed lags of past inflation with the coefficients summing up to 1 (or close to 1). Since the late 1990s, by contrast, inflation dynamics appear to have shifted away from the accelerationist regime, and both inflation and inflation expectations have stayed remarkably stable despite large economic fluctuations (see Figure 3). In particular, following the financial crisis and the recession of 2007–2009, a large unemployment gap opened up that took several years to close. (The unemployment rate rose from 4.5 percent in mid-2007 to nearly 10 percent by late 2009 and stayed above 5 percent until 2016.) The accelerationist Phillips curve 1 Figure 1: U.S. Inflation Source: Bureau of Economic Analysis. 2 Figure 2: Inflation and Inflation Expectations The extended series of professional forecasters’ long-run inflation expectations was taken from the FRB/US model created by the Federal Reserve Board. The FRB/US series is based on data from two surveys: the long-run inflation expectations reported in the Hoey survey of financial market participants from 1981 to 1991Q3 and the median forecasts of long-run CPI or PCE inflation reported in the Survey of Professional Forecasters from 1991Q4 onward, with a downward adjustment of 40 basis points made to the CPI forecasts (all pre-2007 data) to put them on a PCE basis. The extended series of professional forecasters’ short-run inflation expectations is based on one-year ahead CPI inflation expectations reported in the Livingston Survey from 1947 to 1981Q2 and in the Survey of Professional Forecasters from 1981Q3 onward. Data from the Livingston survey were adjusted by Haver Analytics and interpolated to put semiannual data on a quarterly